Where do I find the Audience segments breakdown in Ads Manager?

Click Breakdown in Ads Manager and choose Audience segments under By demographics. If it is missing, set up your segments first under All tools, then Advertising settings, then Audience segments. Meta says the feature may not be available to every account yet.

Is a high existing-customer share always bad?

No. Retargeting and loyalty campaigns are meant to reach past buyers, so a high share there is expected. It matters for the claim, not the campaign. Those purchases are the likeliest to happen without the ad, so test them before you count them as extra sales.

How often should I run the overreport check?

Once a month works, plus after any change to pixels, apps or attribution settings. Use a closed month each time, so late purchases inside the click window have landed, and keep the same steps so the months compare.

How to check if Meta overreports conversions, step by step

Split Meta's purchases three ways in Ads Manager: by attribution window, by first versus all conversions, and by audience segment. Then set each slice against Shopify's first-time and returning customers and confirm each order reaches Meta once. Test the slice that leans on views or past buyers.

By Joris van Huët, Founder & CEOUpdated 2026-10-018 min read

Run the numbers for your store: the free ad platform over-reporting checker.

To check whether Meta overreports, split its purchases three ways in Ads Manager: by attribution window, by first versus all conversions, and by audience segment. Then set each slice against Shopify's first-time and returning customers, and confirm each order reaches Meta once. If a slice leans on views or past buyers, test it.

You need Ads Manager, your Shopify admin and one closed month. Pick a month that ended more than a week ago, so purchases inside a 7-day click window have landed. Use the same dates in every tool.

Step by step

Read the rule behind each ad set. Meta counts each ad set under its own attribution setting, and a campaign total adds them together. Write the setting next to every number you pull. Path: Ads Manager > Columns dropdown next to View Setup > Customize columns > search Attribution setting > Apply.
Split the count by click and view. Put 1-day click, 7-day click and 1-day view side by side for the month. Meta lets you compare them even when an ad set optimises for another setting. The view column is the part GA4 can never confirm, because a view brings no visit. Path: Ads Manager > Columns: Performance > Performance > Compare attribution settings > pick the settings > Apply.
Take out the repeat purchases. The same panel asks for a conversion count. All conversions counts every purchase after a click or view, while First conversion counts only the first. Pull both: the gap is repeat buying credited to one ad. Path: Compare attribution settings > conversion count > First conversion, then All conversions > Apply.
Tell Meta who already buys from you. Audience segment reporting needs two definitions, an engaged audience and existing customers. You pick a custom audience for each, such as a customer list or website purchasers. If you set segments up for Advantage+ sales campaigns, Meta applies them to every sales campaign in the account. Path: Ads Manager > navigation on the left > All tools > Advertising settings > Audience segments (below Creating ads) > Confirm.
Break the month down by audience segment. You get rows for the new audience, the engaged audience and existing customers. It is one of the few breakdowns Meta supports for website purchases. Meta warns the feature may not reach every account yet. Path: Ads Manager > Breakdown > By demographics > Audience segments.
Pull the same month from Shopify. Open New vs returning customers and group by month. Shopify decides who is new from a customer's whole order history, not only the month you picked. Path: Shopify admin > Analytics > Reports > Category filter > Customers > New vs returning customers.
Run the ceiling test. Set Meta's new-audience purchases against Shopify's first-time customers, and its existing-customer purchases against returning customers. A slice that claims nearly every customer in its group leaves nothing for search, email or word of mouth. Path: the step 5 breakdown in Ads Manager, next to the step 6 report in Shopify.
Check that each order reaches Meta once. If the Facebook & Instagram channel and a custom pixel both send Meta a purchase, one order can count twice. Meta's developer guide says it keeps both of two identical browser events. To see what the app pixel sends, click Test next to it and place a test order. Path: Shopify admin > Settings > Customer events > Test, next to the app pixel.
Book a no-ads test for the softest slice. The slice that leans most on views or existing customers is the one least likely to be caused by the ads. Create a Conversion Lift test in Experiments if a campaign qualifies, or exclude a set of regions from the ads. Path for regions: Ads Manager > your ad set > Audience > Locations > hover over the country >... > Exclude cities. If Meta shows Reach more people likely to respond to your ads, untick it.

A worked example

For illustration, here is one closed month for a store, with round, made-up numbers.

For illustration, one closed month	Count
Meta purchases, All conversions	400
Meta purchases, First conversion	360
Meta purchases in the 1-day view column	120
Meta purchases from the new audience	140
Meta purchases from the engaged audience	100
Meta purchases from existing customers	160
Shopify first-time customers	300
Shopify returning customers	500

For illustration, steps 2 and 3 show that 40 of the 400 purchases are repeats, and up to 120 rest on a view. For illustration, step 5 puts 160 of the 400 on existing customers, which is 40% of Meta's count. For illustration, step 7 sets the 140 new-audience purchases against 300 first-time customers in Shopify, about 47%. For illustration, the 160 existing-customer purchases against 500 returning customers come to 32%. Step 8 passes, because only the Facebook & Instagram channel sends Meta purchases.

None of this shows an overcount on its own. It shows where to aim. Meta claiming nearly half of all new customers is plausible if Meta is how most new buyers find you. It is a stretch if search and creators bring most of them. Either way, the existing-customer slice is the softest, so the first test goes on the campaign that leans on it most.

One store's real export adds a warning about the click columns. On its Journeys sheet, journeys of 2 to 3 touches hold 12.2% of revenue and take 12.5 days to buy. If those days count from a Meta click, 7-day click has closed before the purchase, unless a later click reopened it. So in step 2 the click columns can run short on slow buyers, while the view column runs long on quick ones. The export holds no Meta figures, spend or orders, so it cannot score Meta for that store.

What to check when the report looks wrong

No Audience segments option under Breakdown. Do step 4 first. If the option is still missing, Meta has not opened the feature to your account yet, so lean on steps 2, 3 and 8.
Almost everything lands in the new audience. Your Existing customers segment is empty or stale. Rebuild it from a fresh customer list and run step 5 again.
Meta's new-audience purchases top Shopify's first-time customers. Either one order reaches Meta twice, or past buyers are missing from your Existing customers segment. Fix both before you judge a single ad.
First conversion and All conversions are nearly equal. That is fine. Your buyers rarely reorder inside the window, so repeats are not the issue.
Purchases jumped the week you installed an app. Open Customer events and look for a new app pixel that also sends events to Meta.
Meta, Google and email together claim more than you sold. That is overlap between platforms, not Meta alone. The over-reporting checker puts a size on it, and Do Google Ads and Meta double count sales? explains why.

What to do this week

Build the overreport sheet for last month. Give one row each to All conversions, First conversion, the 1-day view column and the three audience segments, then add Shopify's first-time and returning customers. Pass: every Meta slice sits below its Shopify match. Fail: a slice tops its match, so clear duplicates and fix segments before anything else.
Refresh your existing-customer audience. In Shopify, go to Customers, click Export and choose All customers. In Ads Manager, open Audiences, select Create audience, then Custom audience, then Customer list. Check your legal basis first, as Meta asks. Pass: the Existing customers segment uses this month's list. Fail: it still points at an old list, so step 5 undercounts warm buyers.
Put a no-ads test on the calendar. Pick the campaign whose purchases lean most on views or existing customers. Pass: a lift test or a region exclusion has dates before the month ends. Fail: nothing is booked, so the softest slice stays unmeasured.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Frequently asked questions

Where do I find the Audience segments breakdown in Ads Manager?
Click Breakdown in Ads Manager and choose Audience segments under By demographics. If it is missing, set up your segments first under All tools, then Advertising settings, then Audience segments. Meta says the feature may not be available to every account yet.
Is a high existing-customer share always bad?
No. Retargeting and loyalty campaigns are meant to reach past buyers, so a high share there is expected. It matters for the claim, not the campaign. Those purchases are the likeliest to happen without the ad, so test them before you count them as extra sales.
How often should I run the overreport check?
Once a month works, plus after any change to pixels, apps or attribution settings. Use a closed month each time, so late purchases inside the click window have landed, and keep the same steps so the months compare.

Go deeper: Incrementality testing, explained.

Sixty-second versions of these ideas: Causality Engine on YouTube Shorts.

Keep reading

Terms in this article

Browse the full glossary